Date: Tue, 5 Jun 2001 09:15:20 +0800 From: David Xu <bsddiy@163.net> To: Jonathan Chen <jonathan.chen@itouch.co.nz> Cc: Matt Dillon <dillon@earth.backplane.com>, John Polstra <jdp@polstra.com>, stable@FreeBSD.ORG, <chad@DCFinc.com> Subject: Re[4]: time_t definition is worng Message-ID: <83608845.20010605091520@163.net> In-Reply-To: <20010605094735.C40392@itouchnz.itouch> References: <200106012015.NAA17134@freeway.dcfinc.com> <200106012052.f51KqBT29871@vashon.polstra.com> <200106012149.f51LnI289480@earth.backplane.com> <20010605094735.C40392@itouchnz.itouch>
next in thread | previous in thread | raw e-mail | index | archive | help
Hello Jonathan, Tuesday, June 05, 2001, 5:47:35 AM, you wrote: JC> On Fri, Jun 01, 2001 at 02:49:18PM -0700, Matt Dillon wrote: >> : >> :No, time_t is and always has been 32 bits on all FreeBSD platforms. >> : >> :I agree with Matt that it would be nice if it were 64 bits at least on >> :64-bit platforms. Unfortunately, practically speaking it's too late >> :to make that change now. >> : >> :John >> :-- >> : John Polstra jdp@polstra.com >> >> Not being a user of the alpha port, I can't disagree. But if I were >> I would consider fixing the alpha port now and get it over with >> rather then later. time_t was one of the few things that transported >> naturally to 64 bit platforms, and yet it still managed to get fracked >> up in FreeBSD. This IA32 change from long to int only makes things >> even MORE fracked up then they already were and is a huge mistake. >> >> Rather then further break the IA32 port, if consistency is a goal then >> the alpha port should be fixed instead. JC> When the FreeBSD-alpha port was being done, Satoshi (IIRC) suggested JC> that time_t be made a long for the Alpha; which was a great idea to fix JC> the Epoch-end bug, at least for 64-bit systems. After some discussion, JC> this was not done 'cos Terry Lambert pointed out that the UFS filesystem JC> code depended on a 32-bit time_t. JC> I have no idea just how true this is today, but if it still is so, time_t JC> has to be 32-bit until the UFS code is fixed. JC> Cheers. how about userland time_t becomes 64bits and kernel UFS still has 32bits time_t? this is safe until year beyonds 2038. I believe UFS 32bits time_t will be fixed before 2038 :). this way we needn't applications be recompiled when UFS 32bits time_t changed to 64bits. -- David Xu To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?83608845.20010605091520>